]\gt ; the Absorption Spectra of Dilute On the Absorption Spectra of Dilute Solutions By THOSE .
EWAN , B.Sc. , Ph. D. , 1851 Exhibition Scholar Chemistry in the Owens College .
Communicated by Professor H. B. DIXON , F.R.S. Received April , 1894 .
The changes which occur in the colour of a solution when its concentration is chan , are closely connected with the simultaneoub changes which occur in bhe molecular condition of the dissolved substance , and a study of these changes ofcolour may be expected therefore to furnish valuable information as to the molecular ondition of dissolved body .
In a paper published in the ' Philosophical Magazine ' for 1892 , I described some experiments on the absorption spectra ) of solutions of copper salts , which were made from this point of view .
It was found that with increasing dilution the spectra of the different salts examined tended to become the same .
The solutions experimented with were , however , tolerably concentrated , which makes the interpretation of the results difficult , as , according to 's rule , the absorption spectrum of a substance varies to sollle extent with the dispersive power of the solvent in which it is dissolved , therefore , when working with strong solutions there is always the possibility a change in the absorption , produced by changing the concentration of the solution , may be due to the alteration in dispersive .
In order to avoid this uncel.tainty , I have confined my attention almost exclusively to dilute aqueous solutions .
In such solutions\mdash ; differing very little from pure water\mdash ; a change in coluur can only be ascribed , according to our present knowledge , to some change in the ature of the dissolved substance , and therefore definite information as to the latter may be obtained .
A very large number of observations on the spectra of dilute solutions of salts of the same metal with different colourless acids and of the same acid with different colourless metals have been made by Ostwald ( ' Zeit .
Phys. Chem 9 , 579 , 1892 ) .
He as Jred the position of the absorption bands in the spectra , and also photographed them .
Both methods of observation showed that the absorption spectra of all salts containing the same coloured metal or acid were identical .
This is a necessary consequence of the lectrolytic dissociation theory of Arrhenius . .
at the concentrations employed , the salts being almost complete ] dissociated into their , the colour of all the salts containing the same coloured metal , for example , will be due to the same ion .
As , however , uo photometric measurements of the absorption spectra of dilute solutions have been made from this point view , I thought it not superfluous to make a series of measurements Observations of a similar nature have led Magnanini* to express opinion that the absorption spectrum of a substance in solution is to be regarded as independent of the extent to which it is electrolytically dissociated .
A more probable explanation is , as Ostwald has pointed out , that the absorption spectrum of the undissociated molecule very nearly the same as that of the ions , and therefore a change in the degree of dissociation will produce little alteration .
in the spectrum .
A case in which the absorption of the undissociated molecule considerably from that of its ions was also investigated by Magnanini .
Solutions of the salts of violuric acid , in which the salt is almost completely dissociated , possess a violet colour ; Magnanini found the solutions of the free acid to be nearly colour less , although some 8 per cent. of the acid was dissociated into its ions .
J. Wagner , however , did not succeed iu obtaining colourless holutions of the acid .
He found that they had the same colour as the soIutions of the salts , and a comparison of the intensity of colour in a solution of the free acid and in one of the sodium , showed that the intensity of the colour was nearly proportional to the number of ions present .
In dinitrophenol I have found another case of this sort .
The solution containing only undissociated molecules is almost colourless , while ions containing molecules bright yellow .
From the measurements of the absorption spectra of these solntion I have been able to ] the fraction of the dinitrophenol ciated into its ions at different concentrations , and the numbers so obtained are in very satisfactory agreement with those calculated from the electrical conductivity .
Besides the cLanges of colour which may be ascribed to the electro lytic dissociation of the dissolved substance , changes of colour consequence of its hydrolysis by the water .
As an example of this kind of change , I have studied the ferric saIts , especially ferric The results are , however , more complicated than those obtained Rendic .
Lincei , Magnanini , Rendic .
Lincei , Mii Zeit .
Phys. Chem. Absorption Spectra of Dilnie Solution bodies which are only ] issocia , , and it will be bettel to consider them along with the experimental numbers .
Changes of colour may also be due to the breaking up of polymeric molecnles into smaller ones .
It was found that solutions of colloid hydroxide , obtained by dialysis , less light than those obtained by the hydrolytic decomposition of ferric chloride , and in this case there is some reason to suppose that the dialysed hydroxide consists of ] alger moleculal groups than that formed in the very dilute solution of ferric Apparatns Employed .
The measurement of the quantity of of giveu wave-length which is absorbed by a substance in very dilute solution , is not very easy .
In order to compensate for the small absorbing power of the solutions , it is necessary to employ long layers of them , and this makes it almost impossible to use a spectrophotometer such as that of Vierordt .
The appal.atussho } in was therefore put together .
* Both for advice and assistance in its construction , and for the loan of able apparatus , I am much indebted to Professor A. Schuster , F.R.S. thanks are also due to the Grant Committee of the Royal Society , wl1o provided me with the meanf ; of obtaining several pieces of Fig. 1 is a diagramatic epresentation of the arrangement of the parts of the apparatus as seen from above .
It consists of a collimator , wibh two lenses , and , of 20 in .
ocal length , which are both filled with the slit source of light was an incandescent " " Auex\ldquo ; lamp ; and in order to distribute the light mol.e evenly a plate of ound glass was introduced between the lamp and the slit .
When it was desirable ( as in the violet and extreme red ends of the spectrum ) to obtain a stronger tion this was removed .
allel beam of light from the lens passes through the ion ; that from was used as the standard of compal'ison ; the of using only one source of light being to keep the ratio between the tness of these two beams .
The solutions were contaiued in tubes which rested on -shaped supports capable of sliding along the bar , which was clamped to the board on which the apparatus rested , and which allowed of a horizontal adjustment of the position of the tube .
The -supports allowed of a tical adjustment .
The ends of the tubes were ground as nearly as possible parallel to each other , and closed by plates of glass which were simply pressed against the ground ends , no cement or washer being necessary .
The tubes were mm. * A description of a vel.y similar instrument by Lummer and Brodhun appeared in the fur Instrumentenkunde , ' 1892 , p. 133 , when my apparatus was nearly oompleled .
VOL. LVII .
T. .
On the Absorption Spectra of Dilute Solutions .
internal diameter , and the longest ( length metrs ) was provided with five rings of ( 2\mdash ; 3 mm. wide ) , distributed at equal dis{ances along its inside in order to prevent reflexion from the walls .
The shorter tnbes were provided with black paper rings at their ends for this purpose .
After passing through the solution , the light is reflected internalIy by a right-angled flint-glass prism , and falls on the Lummer and Brodhun prism The construction of this photometrio prism has been described minutely by Lummer and Brodhun ( ' Zeitschrift fur Instrumentenkunde , ' 1889 , p. 461 ; and 1892 , p. 46 ) .
It consists of two -angled prisms , ground and polished together on their hypothenuse faces , and held together as shown in fig. 2 .
Certain parts of the surface of one prism are removed , at for example , so that the light coming from is totally reflected at the sulfaces , but transmitted at .
The light from lens is on transmitted at the surfaces .
Consequently , an eye placed at sees the surface , illuminated on the parts marked by Iight from on the parts by light from the lens .
The measurement then consists in varying the brightness of the light from Jens , until the brightness of the different surfaces is the same .
By means of glass plates at ( fig. 2 ) , a reading by equal contrast between pairs of surfaces of somewbat different brightness , can also be made at the same time .
For the of this arrangement , however , I must refer to Lummer and Brodhun 's original papers .
The variation of the intensity of the light from lens was prodoced by means of the rotating sector described by Abney ( ' Phil. Nag vol. 27 , p. 62 , 1889 ) .
The sector was made by Mr. A. Aiter passing through the prism , the two beams of light follow the samo path through the two rless { lint-glass isms G and I ( one of , the other ) .
The lens then throws the spectl.um thus pro- It was made by .
Schmidt and Haenscb , of Berlin .
Absorption Spectra of Dilute Solutions .
glass plate on which the prisms stand was then placed as nearly as possible horizontal , and the prisms adjusted with their refracting edges vertical by reflecting the light from the slit , , first from one surface and then from the other inbo the telescope , ( the slit of this telescops being replaced by an eye-piece with cross threads ) until the images eflected from the two surfaces were seer in the same position .
The eSCOPe , , was then placed horizontal , and tho collimator levelled , so that the images of the slit , , seen in the telescope were in the centre of the field of view .
The right-angled prism , , was then put in position and levelled , so that the light underwent no vertical displacement on through it .
The Lummer and Brodhun prism is next treated in the same way , and finally the two refracting prisms are put in position and adjusted to minimum deviation .
The at which the light was internally reflected in the right-angled prism was in my apparatus .
As the critical angle for flint glass is there was no danger of the internal refle-xion being incomplete .
Finally it was found to be of great importance that the light should pass centrally through the tube of solution .
To adjust the tube , cross threads wsre attached to the lens , .
This was conveniently done by means of a brass ring which just slipped over the tube carrying the lens to which the threads were attached .
A card with a pin-hole in it was set up at a distance of about a metre and half from the lens in such a posi- tion that the prolongation of the line joining centre of the slit , , and intersection of the cross threads passed through the pinhole .
The tube of solution , also provided with cross threads at each end , was then placed in position and adjusted , so that , on placing the eye behind the little hole in card , the intersections of all threo cross threa feli together .
It was fouud that a perfectly satisfactory adjustment could be got in this way .
Method of the The sector being open to its full extent , the brightness of the lights was arranged by pntting a smoke-glass plate in the collimatol at , so that the comparison light was alittle brighter than the othel The width of the slits was taken as small as was consistent having sufficient light .
Five readmgs then made of the ratio between the intensities of t , he two beams .
The tube of solution then put in position , the two sted to coincide , and { iv more readings made .
Then the readings were repeated , the tube being , however , the second time placed in such a position that defleotiou of the light was in the opposite direction ; consequently the tube rotated round its long axis .
When ths means of these sets of readings did not crree , the adings of average oodness , and will serve to show the sort of differenc which occurred between the single readings .
The numbers given the openings of the sector .
eduction of Results .
The quantity which is directly measured is the amount of lighi which is lost by passing through a system of two plates with a layer of aqueous solution between them .
In order to calculate from these numbers the part of the light absorbed by ffil dissolved substance , it is mecessary to allow for the light reilected absorbed by the glass plates and the water .
To obtain the formulae , I have applied the equations given by Stokes ( ' Roy .
Soo , Proc vol. 11 , p. 545 ) to special case as follows:\mdash ; Call the fraction of the incident on the glass plate from air , which is reflected , and the same quantity when the light is iocident from the liquid on the glass plate are the refra , ctive dices for air to glass and water to respectively .
And suppose that a fraction of the light 1 absorbed by passing once through the glass plate .
Then , allowirl for the repeated reflections between the two surfaoes of the plate , th quantity of light ( calling its ( intensity ) which will into the liquid is pectra of Solnfions .
and the pal.t reflected will be If the light is incident on the plate the liquid instead of air , the part transmitted will be the same , while that reflected be Now consider in the same way the layel of liquid as being a plate , the end surfaces of which have the property of transmitbing a fraction of the light incident upon them , and of parts or nccording as the light is incident externally or internally , and which uxther transmits a part , , of the light which passes through its whole once .
Take to be the transmission coefficient for the solvent , and that for the dissolved substance .
Then the part of the light transmitted by the whole system is . . . . . . . . . . . . . . . . . .
( 1 ) .
is the quantity is directly ; from it we have to calculate ) , ) are determined by nary expts .
The values of and depend on the refractive indices of the glass plates and of the solutions .
The used were all so dilute that no appreciable made by using the same values for anl as are for found the value of , the extinction coefficient .
the dissolved body , is obtained by means of the The calculation of from tion ( may bc much simplilied in the following .
We write equation ( :\mdash ; the glass plates used the greatest ) ossible value of was .
So that approximately , from which , ouly depends on -length of the nsed , and is most conveniently obtained from a drawn once for all .
Also is following table .
In it is the light transmitted by the two nnd transmission coefficient for mm. of the glass:\mdash ; .
of Dil , Solntions .
possible from dust , ordinary distilled watet ' and distilled a second time from a metal can with block-tin condc.nhel , the first third of distillate being rejected .
This water , filled immediately into the tube , was almost free from vi , ibledu , t in a layer 1 .
long .
It was just ibly bltle .
I iound that water filtered tbrough " " Chamberland Pasteur candle \ldquo ; was even more devoirl of dust than that which had been frej , tilled , and some were made with such filtered ater .
The ] vfis , , no noticcable diIference in the The actual measurements of the mitted I ) ' a tube of vater 1 cm .
long aro given in the I. A curve smoothed as Iittlc as wa drawn these and the values of in II interpolated from it .
the values of the extinction coefficicnt for 1 cm .
ater ( calcolated .
I I. .
fncr The umbers taiued } ) Hufn .
and ( ' .
Ann 42 , p. 10 , 1891 ) are luded iu table .
The factor exccpt at ths .
Owing to thit , lack eement I repeated the measurementf\ldquo ; in part of spectrum t , everal time } , but never found fner and Albrecht .
In fig. the valncs of , are plotted the .
The abwrption diminishes the olaI line A and Dr. .
Ewan .
On the FIG. 3 .
and again the neighbourhood of , corresponding to the two absorption bands which have been by several authors .
* It is also noticeable the absorption slightly increases in the violet near G. Solutions of Cupric Salts .
used were cupric sulphate , nitrate , cbloride , bromide , and acebate .
They were purified cal.efully , notwithstanding which the solutions made by dissolving them in water were not sufficiently clear to be measured .
Filtration through fine Swedish filter paper only improved them a little .
The eedom from suspended particles was , howevel .
, easily obtained by filtration through pol'ous earthenware .
The apparatus shown in fig. 4 was found to be very venient , as it allowed of two or three litres of tion b without requiriug ally attention .
A and are ordinary Winchester quart bottles , is the " " Chamberland Pasteur caudle The solution to be filtered is contained A. is evacuated , and the apparatus left to itself till the solution has fiItered .
Two litres could be filtered in an hour , or less if a good vacuum was kept up : The solutions measured were always analysed as the filtration changes their concentration somewhat .
Two methods of analysis were used , and the mean result taken .
* Schonn , ' Pogg .
Ann. Ergz 8 , p. 670 , 1878 ; Russel and Sapraik , ' Natul.e , ' 22 , ' 1880 ; and Soret and , ' Arch. de Sc. Phys. et Not vol. 11 , p. 337 , bsorption Spectra of Ditute Solufio ( 1 .
) From to llitl.e of the solution was evapo ) ated to dryness ( with a liblle hydrochloric or acetic acid ) on the water bath , tho residue dissolved in a small quantity of water , excess of potassium iodide added , and the liberated iodine titrated with sodium thiosulphate solution .
In order to obtain , ccurate results by this method the cupric solution must be concentrated , ] lerwise the reaction with potassium iodide is incomplete , and the thiosulphate solution must be standardised by means of pure in ctly the same way as it is subsequently to be used in the analyses .
( 2 .
) The solutions , without eval ) oration , were ated with potassium cyanide solution in presence of ammonia .
Here also the KCN solution was always standardised by means of a solution of of known strength before using it , as ibs strength on keeping .
The results of the two methods The resuIts of the photometric ements are contained in Table II Appendix .
In order to be able to compare tlJe absorption spectra of the differeut solutions , the extinction coefficients for the dissolved substance for 1 cm .
layer , and for a concentration of 1 gram molecule perlitre , have been calculated from the values of given by equation 2 ( p. 125 ) .
The extinction coefficient is proportional to the length ( l ) , and to the concentration , so that From thefie values of curves were drawn ( not smoothed ) , and from them the numbers contained in the following table were interpolatod .
The Roman numerals at the head of each columu indicate that the figures are derived from the experiment similarly numbered in the Appendix .
Dr. T. Ewan .
) Absorption Spectra of Dilute Solutions .
In some cases the numbers given by two different sets of measurements were so nearly the same that they were united to one curve .
This was the case with III and II and part of the numbers for ; and also with the two solutions of and XI .
Leaving for the } ) the numbers for out of account , it is ident that the differences which occur between the numbers obtained with different solutions of the same salt are as great as those which are found between solutions of different salts .
The absorption spectra of the four salts , and must accordingly , in the concentrations used , be regarded as the same .
In the following table ( p. 132 ) the means of the above numbers are compared with the numbers which I formerly obtained ( ' Phil. lfag .
' [ 5 ] , vol. 33 , p. 317 , 1892 ) .
These numbers show that the absorptions change very little when the solutions are diluted .
The electrolytic dissociation , however , challges considerably .
For the fraction of the salt ally dissociated increases from in the stronger soIution to in the more dilute .
For the increase is from to That the change in dissociation produces very little change in the absorption is most simply explained by supposing that the Cu existing in chemical combination with other atoms in the soIution has the same absorption spectrum as the copper ion .
In the case of this leads to a result some interest .
For the concentrated aqueous solution of cupric and also its olic and ethereal solutions are brownish-green colour , and as these solutions doubtless contain the molecule we cannot ascribe to it the same absorption spectrum as to the Cu ion .
In the dilute solutions of examined , therefore , the comb ned part of the copper probably exists in the form of the ion .
This is aJso in good agreement with ] the observation which I made formerly ( ' Phil. bIag 1892 ) : that the rapid change in the absorption spectrum which accompanies the dilution of a conccntrated solution of ceases when the concentration reaches a value ] ying between and gram molecules per litre .
The fractions of the salt dissoci these concentrations , calculated from the electrical conductivities of the solutions , are and respectively .
It wouId appear thus that on diluting a strong solution of chloride the molecules first dissociate into the and Cl , and that the dissociation of the ions only takes place to any considerable extent after this process is tolerably complete , that , after the dissociation of the salt has reached 50 cent. This behaviour is very common the case of dibasic acids , of which a large number have been examined by Ostwald Zeitschrift fur Phys. Chem vol. 3 , pp. 186 and 280 , 1889 ) .
It will be seen from Table III that the extinction coefficients of Dr. T. .
On the Absorption Spectra of Ditnte Solntions .
solutions of copper mtlcl ] larger than thosc of the other The of two solutions , emonts on are there given , and gram molecules per litro respectively .
The ( Table V ) that tho extinction coefficients diminish rapidly when the conccntration becomcs smaller .
The most dilute solution used still gives largel extinction coefficients than those found for the .
coppcr salts , but it unfortunately not possiblo with ] my to accurate measurements with still smalIer tions , so that 1 to find out whether the absorption spectrum of copper becomes ultimately tical with of tLcothc.rs or not ; it appears , howevcr , to tcnd to do so.* .
Two explanations of the behaviour of copper tate solutions themselves : The electrolytic dissociation into Cu ions may be incomplete , the still undissociated part of the salt may have a different colour from the Cu ion ; the salt bo partially ydrolysed , acotic acid being a so much weaker acid the others .
In the latter case , however , ons wouId expcct difference betwee1l the spectrum of copper acetate and that of the other salts to become grcater with increasing dilution , as in the more diluto solutions the bydrolytic decomposition would bo most advjlI ) The opposite , however , is tho case .
Tbo addition of acctic acid to the solutions should , by amount of hydr.olysis , bring the copper acetate spectrum neal .
Cl. to that of the othels , if tho Tho uncertainty of observations of bsorption s of pbotometric meastlrements is illustrated by the following remark of Wied Ann vol. 4$ , p. 754 ) .
He says : " " Ebensowenig wie bei rioh bci ) cctut mit der Verdunnung cino Veranderung im Character dcs pectrums .
Beobachtung der Absorptions-Grenr .
im Roth bei diosem alzo fur verdunnto Losungen durch dio Absorption des Wasscl .
beein- htigt Table V shows that not very considerable dilution nbout to flve ) roduced the absorbing of il copper solution by about ono-hadf .
of Dilute Sotutions .
Table Conductivity of Copper Acetate Solutions .
At vol. in in which 1 graJll molecnle is dissolved .
molecular conduotivity . .
aotion of coppel .
acetate dissociated ions .
seems probable then that just as the dibasic acid spht off one hydrogen atom as an ion and then the second , so the copper acetate splits up first into the ions Cu and , and then the ion breaks up , after the first process is ] complete , int the ions Cu and .
] atedfi.om the numbers for the electrolytic couductivity given above the quantities of the salt sociated in the two solutions in Table are\mdash ; Conc . . . . . . . . . .
cent. , , ( extrapoIated ) .
These two solutions will accordingly contain the ions Cu and , the gl.eater the quantity of the lattel .
the more will the colour of the solution differ from the normal .
The addition of acetic acid to the solution diminishes the dissociation of the copper salt , and increases the difference of the colour from the normal .
Salts of Dinitrophenol , .
The dinitrophenol used was the ordinary one ; the specimen obtained from Kahlbaum melted at .
The potassium and ammonium salts were prepared from it .
They both crystallise veryv well , the .
in orange , the latter in bright yellow needles , and are therefore easily obtained pure .
The potassium salt lost on heating to per t. Calculated for VOL. LVII .
Dr. T. Ewan .
On the The ammonium salt volatilises on heating .
therefore determined .
found .
, . . . . . . . . . . . . . . . . . . . . . . . . . .
per cent. calculated for . . . . . .
As I could find no satisfactory way of analysing the dilute of hese salts , I made them by dissolving a weighed quantity of carefully purified saIt in water .
The solutions so obta , ined were fairly clear .
The values of ( the extinction coefficient for the dissolved body for 1 cm .
layer and 1 gram molecule perlitre concentration ) are given in the following table .
The experimental numbers are given in the same as before in the Appendix Table III .
The numbers given in Table VIII are obtained from the number in Table III of the Appendix interpolation .
The concentrations are in gram molecules per litre .
The two most absd solutions used give somewhat smaller llumbers than the hers , which may probably be attributed to the electrolytic dissociation being incomplete in the stronger solutions .
The numbers obtained with the othel solutious vary irregularly ; by taking mean the following numbers are obtained .
salt .
salt .
The absorption increases so extremely rapidly between the wave lengths 470 and 500 that a small error in the measurement of wave-length can make a great difference in the value of obtained .
Taking this into account the absorption spectra of the potassium and ammonium salts may be regarded as the same .
These salts sociate electrolytically into the ions and The mean of the above numbers therefore be as represent ing the absorption spectrum of the ion The behaviour of the solutions of the free is of interest .
The solutions in benzene or be the are brownish-yellow , the colour is faint , and unless they are fairly concentrated thoj appear colourless .
An alcoholic solution is much more Absorp tion of Dilute Sotutions .
obtain the spectrum of the undissociated dinitrophenol a solution of diuitrophenol to which excess of hydl.ochloric acid had been was used .
The dinitrophenol had been recrystallised from water , notwith .
standing which the solutions obtained from it were never sufficiently clear .
They had , therefore , to be filtered rough the porous earthenware and analysed .
The analysis } made by titration with dilute ammonia solution , using litmus as indicator .
The change of colour not being very sharp , standard hydrochloric acid was added when liquid was acid , and then ammonia till alkaline , and so on several times .
Tests with solutions of known concentration showed accurate results could be obtained in this way .
The following table contains the extinction coefficients of the soIution containing cxcess of acid , and also rison , the mean of the numbers obtained with the potassium and ammonium salts .
Table IX .
It is evident that the ption of the acid solution is almd compared with that of the solutions of the salts .
With the help of these numbors it is possible to the * Arrheuius , chrift Phys. Chem vol. 2 , p. 284 , , and vol. Absorption of Dilute Solutions .
tion of the dinitrophenol which is electrolytically dissociated in a solution , a measurement of its inction coefficient for any one wave-length having been made .
If is the part is dissociated into its ions , the extinction coefficient of the solution ( calculated for nnit concentration and 1 cm .
) will be Table X contains the resuIts of the measurements of a numbel of solutions of dinitrophenol of diffe ] concentrations .
For the mental details see Appendix , Table .
In Table X , , and are the extinction coefficients for the solution actually measured , for the ion , , and for the undissociated molecule of phenol respectively .
The values of , as remarked above , are taken as the mean of the measul.ements made with dilute solutions of the potassium and ammonium salts .
the fi.action dissociated is lated from the equation given above .
Table X. The numbers under the fiactions of the substance electrolytically dissociated ated from the determinations of the electrical conductivity of solutions of dinitrophenol made by Better .
( ' Zeit .
Phys. Chem vol. 6 , p. 298 , 1890 ) .
He found the dissociation oustant of dinitropbenol .
This is connected with by .
V , the volume in litres in which 1 gram ] ecule is dissolved , may be replaced by , which gives Dr. T. Ewan .
On the The agl.eement found between the numbers obtained Srom two il very different properties as the absorption of light and conductivity , is so striking , taking into account the ver7 considerau experimental error , doubt can be as to the substanhI accuracy of the hypothesis on which the whole calculation is based .
Salts .
The changes of colour which take place when the aqueous solution of a ferric salt is diluted are so considerable that they appear to offer a very suitable means of determining the amount and nature of the decomposition which the salt undergoes .
In 1856 , Pean de St. Gills , *made a study of the decomposition which ferric salts undergo on heating their aqueous solutions .
lloticed the peculiar appearance of the solutions containing fernc hydroxide , they being clear by transmitted , bnt apparently turbid by reflected light .
He was aware that the hydrat , exists in the liquid in a state of pseudo solution and uncombined with the acid present , as the following remark Il semblerait plutot que l'acide acetique exerce sir l'hydrate un action purement mecanique en divisant a l'extreme les particules iusolubles et simulant ainsi un dissolution veritable In 1861 , Graham showed how the colloidal ferric hydroxids oam be prepared by dialysis , and described its principal properties .
He regarded it as being diffel.ent from the soluble hydroxide obtaino4 by Pean de St. Gills , by heating the solution of ferric acetate .
in 1869 , Tichborne , S 1871 , and Berthelot in 1873 , made studies of decompositions which occur both on heating the soIutions and on diluting them with wate .
Their results , however , do llot lend themselves to a calculation of the amount of which has taken place .
In 1871 , Kreckel made a number of minations of the quantity of ferric hydroxide formed in solution8 ferric chloride of different concentrations , and at various The amount of ferric hydroxide present determined by ing it with sodium cbloride .
He found that the precipitate formed bj heating solutions of ferric chloride containing than 4 per of the salt contained a variable amollnt of chlorine .
The mon * ' Ann. Chim .
Phys ( 8 ) , vol. 46 , p. 47 , 1856 .
'Phil .
Trans 1861 , p. 209 .
Aanteekeningen , etc. , Prov .
Utrechtsche Genootschap , 1869 , p. 14 .
S 'Roy .
Irish Acad. Proc 1871 , p. 169 .
( Ann. Chim .
Phys ( 4 ) , vol. 30 , p. 145 , 187$ .
' J. prakt .
Chem , ' ( 2 ) , vol. 3 , p. 286 , 1871 .
bsorption Spectra of Dilute Solutions .
dilute solutions deposit only ferric hydroxide .
He found also that the amount of decomposition increases with rising temperature and increasing dilution .
Finally , G. in 1878 , ) tlblished an investigation on the hydrolysis of salts , the delerminations of the amount of hydrolysis which had taken place being made by means of the moIecular netism of the solutions .
The molecular mag.netism of iron the fo of hydroxide being of its value in a strongly acid solution , a meastlrement of the evnetism of a solution of known strength allows the ratio which the iron is divided between the two forms to be ] culated .
The fact that , as ecke observed , the precipitate formed by heating solutions of fell'ic chloride contains chlorine , and also an vation m by Picton and that on filtering a solution of colloid hydroxide chlorine through a porons cell the filtrate contained neither iron nor chlorine , led me in the first place to attempt to whether bhe roxidp formed in dilute solutions of ferric chloride contained chlorine or not .
this purpose solutions of known concelltration were filtered through a porous ocll .
The cell was first well washed distilled water , then a of the solution filtel.ed through it to remove water from its pol.es .
The nexl portion of the filtered solution was collected and aualysed .
The difference between the composition of the original solution and that of the filtrate gives the composition of the substance held back by the porous cell .
The hltration apparently removed the , completely from the soIution .
The yellow-brown coloured solutions were , after filtration , either completely colourless or very faintly yellow , to their concentration .
This method of analysis has the drawback that the solution changes its concentration to Home extent during the filtration partly owin , to ation in the vacuous vessel into which it is filtered , parfly owing to difficulty in removing the water which has served for washing the cell from its pores .
A direct analysis of the precipitate is hardly possible as it cannot } ) collected on a filter , nor oan ib be Yashed by decantation , as it refuses to settle .
Further , to wash it with water is not allowable , as it would tend to change its composition .
The ferric chloride used was purified by repeated reof the hydrate , an analysis of its solution showed that it contained iron and chlorine in the ratio sented by its formula .
* ' Wied .
Ann vol. 5 , p. 45 , 1878 .
'Journ .
Chem. Soc 1892 , p. Bakhuis Roozcboom , ' Zeit .
Phys. Chem vol. 10 , p. 477 , 1892 .
From which ratio Ratio calculated for Two solutions which were subsequently analysed , tained grams .
Fe per litre to rams e. Cl. and , , , , The dilute solutions made by diluting a concentraed solution of known concentration , were allowed to stand for some weeks at the ordinary temperature , so that a condition of equilibrium was almost certainly reached .
The following table contains the results of the filtration experi .
ments .
The concentrations are here ( as elsewhere in the given in gram molecules per litre , the molecule of the salts being always taken , for convenience , as the quantity containing one atom of iron .
Table XI .
The decrease in the concentl.ation of in the solutions , caused by filtration , is so small ( in one case was even an increase ) , that it appears probable that the substance separated from the solutions is really a hydroxide of iron , and contains no chlorine .
changes of concentration the chlorine are probably sufficiently explained by the errors inherent in the method used , which havs been already mentioned .
On the othel hand the loss of iron inCl.eases as the solutions become more dilute .
Walker*found , in an investigation on the hydrolysis of the salts of some weak bases , that the decomposition took place in agreement * ' Zeitschrift .
Phys. Chem vol. 4 , p. 319 .
1889 .
Absorption Spectra of Solutions .
with the law of Guldberg and Waage .
This is , however , here not the case .
The ) rium which takes place dilute solutions of ferrlc chloride may be most simply represented by the equation Suppose that to with one gram molecule of was di solred in molecules of water , neither hydrochlorlc acid nor ydroxide being present .
When the equilibrium is reached , ) pose that a fraction of the ferl chloride molecules have decomposed .
Then we have molecules molecules .
molecules , and molecuIes of water .
As is compared with , this may be taken as the same as According to Guldberg and Waage these quantities should be connected togethor by the equation .
( 3 ) .
is identical with the loss of iron given in Table XI divided , and , calling the concentration of the original unfiltered solution ( with regal.d to Fe ) , we have 1 gram molecule in , as llitre of water tains molecules , we have 1 molecule in ] ecules H. Putting these values in the above equation we get The values of , calculated from the values of given in Table XI , and are evidently very far from being constant , as is required by the law of Guldberg and The eement is hardly improved by arding the ferric hydroxide us an insoluble body and putting ibs active mass constant .
The equation then becomes cons The values of also vary considerably , as Table XI shows .
According to the equation for the equilibrium of four electrolytes , on the right-hand side of equation ( 3 ) , is not a const , ant .
rhenius rftrds only parts of the electrolytes are dissociated into their ions as the active masses , so that if , and are the fract-lons of the electrolytes , and HC1 respectively , which are dissociated , must substitute for the equation of Guldberg and Waage the expression ' Bihang .
Svensk .
Vet.-Ak .
Hand vol. 8 , No. 14 , 1884 .
Dr. .
On the As Arrhenius* shown , the values of , \amp ; c. , are to be tained from equations of the form . . . . . . . . . . . . . . . . . .
( ) .
the whole number of dissociated molecules of all kinds in solution , and , \amp ; c. , are the characteristic dissociation constanh for four electrolytes .
The es of and are very small , so that have and , the fraction of the hydrochloric acid dissociated , may be put , so that the equation , putting . . . . . . . . . . . . . . . .
( 6 ) .
ecting small quantities , we have Substituting which in ( 5 ) , . . . . . . . . . . . . . .
( 7 ) .
The values of , the dissociation coefficients of ferric chloride , are unknown .
I have attempted to test the agreement of the ments with lihe Equations 6 and 7 , thus :Each experiment gives a for ( say ) , or , writing the constallt part , we Substituting this in Equation 7 , From these equations and can be found .
The following numbers were obtained:\mdash ; * ' Zeit . .
Chem vol. , p. 1 , 1890 .
Cf .
J. J. van Laar , 'Die dynamik in der Chemie , ' 1893 , where a convenient collection of the be found .
Absorption Spectra of Dilute Solutions .
lIean used Mean The numbers now found are ] constant .
As , we is , dissociation constant of ferric ydroxide is about 9000 times that of water .
Ferric oxide is thus an extremeIy weak base .
we also compariso the dissociation constants of , that of , so that must be regarded as one of the exccptionally slightly dissociated salts .
A number of photometric meaRurements were made of the tion spectl.a of solutions of ferric chloride , but their peculiar properties make it rather difficult to obtain accurate results .
The solutions were made by diluting a concentrated solution of ferric chloride , which had been made as clear as possible by filtration papel ' .
They were then allowed to stand till no further cLange { ook place in their colour ; if they were then withouli isible t , they were used for measurement , but , owing to the peculiar fiuorescent appearance of solutions containing colloid fcrric hydroxide , it was not easy to see whether they were really free solid iCleh or not .
The numbers which contained in Table of the Appendi and from which the following table is calculated , give results which agree approximately with the resuIts of the filtration experiments given ave .
The calculation of the amount of fert.ic hydroxide the solutions was as follows : most dilute solution of which used ( XXV , Table XII ) , and another one to a small quantity of ammonia had been added ( XXVI , Table XII ) , gave nearly the same numbers .
The decomposition in these soIutions appeared to be complete , as neither of them gave a red colour on addition of excess of ammonium snlpbocyanide .
Solutions of colloid ferric bydroxide , obtained dialysis , also give no red colour with anide .
I took the absorptions of these solutions as corresponding to complete hydrolysis .
The smaller absorptions of the other solutions will then correspond to a , tely smaller decomposition , if we assume that the whole absorption is due to the hydroxide , an assumption which is quite allowable , as the filtratioI * The numbers are van Laal .
, .
To convert them into units used by Ostwald they must be divided by Dr. .
iwan .
On the Spectra of Solutions .
experiments , the removal of the hydroxide from the solubions leaving them lrless .
All the solutions in which the hydrolysis was incomplete gave the red c.olour with ammonium sulphocyanide .
The meaSUl.ements which G. Wiedemann made of the amount of hydrolysis in the more ) ncentrated soIutions of fcrric chlo appear be in harmony with the results given in Table XI .
lumbers , when plotted ainst concentrations , give two to be parts of the same curve .
Colloid Hydroxide.\mdash ; A solution was by dialysing a solution of chloride which of droxide had been dissolved .
It was clear by transmitted but apparently curbid by reflected light ; it gave no deposit on for three months , and passed through filter paper without ' any residue .
A solution of coagulated it , but gave no red colour .
alysis showed that it contained per cent. of the amount of chlorine to the formula .
The extinclion coefficients obtained for differcnt solutions , made by diluting this solution , are iven in Table XIII ; they are obbained by graphic interpolation from the experimental numbers given in the Appendix , able V Table XIII .
The , on the whole , show some tendency to increase as the dilution lcreases .
A comparison with the numbers obtained for the completely hydrolysed solutions of ferric chloride ( last olumn , XIII ) shows that the spectra are entirely diffelent .
can be little doubt that both the solutions contain colloid ferric .
The filtration experiments show thab this is the case in a of tartar emetic into o1le of sulphuretted hydrogen , possessed only a faint fluoresc.ence , and , under the microscope , no particles were visible in it .
On dialysing the solution the fluorescen increased , and gradually minute particles became visible under the microscope ; finaily the ] ated .
Picton ar-d Linder J. Chem. Soc 1892 , 154 ) found also that silicic .
acid behaves in a similar way .
a colloid solution of silicic acid containing lydrochloric acid { he particles were too small to reflect light , while after dialysis they were large enough to do so .
Krecke .
cit. also found that , on dialysing a concentrated ferric chloride solution , it finally coagulated .
It would as though the dialysis always favoured the formation of ] molecular aggregates .
A possible reason for this , in the case of ferric hydroxide , at any rate , is not difficult to see .
In the solutions containing hydrochloric acid the droxide must be egarded as constantly being acted upon by the acid with formation of chloride , which is at once decomposed again .
Consequently the hydroxide 1lever has the opportunity of forming large molectllar groups .
ltions ctaining excess of Hydrochloric Acid .
The addition of a small quantity of acid to a solution containing the hydroxide diminishes the quantity of the latter , and therewith the intensiy of the brown colou of the solution , to a marked extent .
I found , for example , by filtration through the Chamberland Pasteur candle , that the addition of one-sixbh of an equivalent of HC1 to a dilute solution of ferric chloride reduced the tity of present by nearly a half .
The following are the .
Absorption Spectra of Dilute So fntions .
La.rger ties of acid seem to prevent the formation of ferric hydroxide altogether , as the acid soIutions could all be filtererl through the porous cell without any deposit .
The analyses also showed that their concentration was not diminished by the filtration .
Table contains the numbers which were obtained with a sel.ies of acid solutions .
The meant of the table is the same as before , only the number of equivalents of force hydrochloric acid added to equivalent are also give ] ] .
Table These numbers appear to be somewhat irregular , but a glance at fig. 5 will show that this is due to the way the curves cross each other .
The curve XXX of the soIution containing the smallest excess of hydrochloric acid has a strong resemblance to the curve for , which is included in the figure for comparison ( with ordinates divided by 100 ) .
If the whole absorptiou of soIution XXX were due to hydroxide , the amount of hydroJysis would only be to per cent. of the salt present .
the curve for the next solution , XXXI , the similarity to that of the hydroxide has disappeared , and as the quantity of acid is increased the curves ntly approach more and more towards that of the concentrated acid soIution XXXV , which may be taken as representing approximately the absorption pectrum of the uttdecomposed molecule of ferric chloride .
Whetho ] these changes are to the change of the dissociated ferric chloride into undissociated , or to the convel'sion of a soluble oxychloride into ferric c , bloride cannot be decided .
It is certain that these solutions none of the colloid hydroxide , but they possibly contain soluble oxychloride .
Wiedemann found that solutions of ferric nitrato and sulphate , containing .
excess of acid , still contain a conDr . .
DWENV .
On the c ) coeJJicients .
Spectra of Dilute Solutions .
siderable quantity of hydl.oxide .
He has given no measul.ements on acid solutions of the oride , but as I found that the acid soIutions of ferric nitrate and sulphate behave very much in the same way as the chloride with respect to colour , we may perhaps conclude that all these solutions contain some reaily substance standing between the undissociatod llolrnal molecules and colloid hydroxide .
Some surements which were made with ferric nitrate and sulphate ' results similar to those obtained with the oride , but they are very milch less complete .
The table contains the measuroments made on concentrated acid solutions of ferrio sulphate and nitrate , and with a solution of ferric sulphocyanide in ; owing to the ease with which the tter decomposes the numbers are only approximate , they will serve , howcver , to give some notion of relative of its absorption .
For comparison the numbers obtained for the conccntrated , acid , ferric solution , and for ferric hydroxide are iven .
Table olutions of erric SaIts .
It will 1 noticed the of iron all have different absorption , the nitrato is slightly coloured indeed ; the sulphate brown , and the ight yellow .
Ferric hydroxido absorbs times tho salts , and ferriu llctrly timas as much light as the hydroxide .
ENDlX.\mdash ; Table I. is the hght tran by a tube full of water cm .
long , as ectly deter .
mined .
Table II .
The letters the ving significations : \mdash ; ation of the tion in gram .
per litre .
The molccular weight used is given fter formula of ) substance .
of solution .
ght )allsmitted b tnbc of solution .
ient for 1 cm .
ot solution -moleculc per litre .
\mdash ; Ir .
cm . .
cm .
Spectra of Jilnte Solutions .
1 Table of Ditute Solutions .
Table III rophenol ( Acid Solution .
molec ulc per litre .
cm .
Dr. T. waJl .
On the Table Table \mdash ; Colloid Ferric Hydroxide .
Table IX .
VOL. LVII .

